(SF driver): the moat’s not taste—it’s attention. the future is agentic agents
(Me): that’s it, I’m done [jumps out of uber, hitting homeless woman]
(Homeless woman): hi! new to sf, here’s an intro—i’m mildred, 2x umich dropout, former @AIatMeta, 1370 SAT, building in public
i have seen enough proof now that using a coding agent is a deep skill
it's confusing because the people you see heavily using them produce horrible results
but that's because it's a skill! you can get better and the ceiling seems pretty high - this is very exciting to me
It’s good to periodically prompt your chatbot about something you know exceptionally well, just to remind yourself that it doesn’t know what it’s talking about.
Last vibe flow worked really well. Yesterday/today got a fastAPI app up on Google Cloud Run. It figured out the session cookie we need from site we want to hit, got docker image from 414MB to 82MB, gave exact commands for me to deploy. Still unreal
@sudo_goreng Writing good tests requires more skill than writing good code. Really. But people don't understand that and so focusing on tests isn't a good career move, you'll be pretty much invisible.
@latentspacepod@simonw@ankitxg "The agent implements. The BDD framework verifies" is doing a lot of work here, we can't have the spec be separate from the program, bc we can't trust LLM to interpret "Then they receive an email with a reset link" correctly and write `sendEmailSpy.calledWith(...)`
it's crazy what a 1.5B model can do these days!
"VibeThinker-1.5B is a 1.5-billion parameter dense language model. With a total training cost of only $7,800 USD, it achieves reasoning performance comparable to larger models like GPT OSS-20B Medium."
runs perfectly on device!